[Reference] Nichia E17A/E21A (2000K - 6500K, R9050/R9080, color) CCT and tint shots

This is a dedicated thread to make choosing/mixing Nichia E17A/E21A easier. Maukka’s measurement data will be added later. We measure/photograph exactly the same LEDs

Work in Progress

Bare white PVC reflector

1 meter from the wall.
All shots used 700mA constant current per LED.


  • Body: Olympus OMD E-M5, [WB5600K, manual mode. RAW].
  • Lens: LUMIX G VARIO 12-35/F2.8

Image processing:
DxO Optics Pro 11

  • geometry corrected per camera body/lens combo with built-in module
  • 5600K manual set white balance
  • color rendering: DxO portrait

Click to ENLARGE

Note: The first 4 pictures the wall looks dry, then the rest of the pictures you can see water falling. There was a huge storm in the middle of the shoot.

Edited version for better viewing. The processes goes like below:

- Partially cropped near the lower left center from the cross

- Gaussian blurred twice

- Exposure of each shots adjusted to match nearby shots


Finer brightness tuning plus some more info added:

E17A Color mixing (Red, Green, Blue, Amber)

White matte PVC LED holder only

20mm Plastic aspheric, 15mm focal length

20mm 10° polished surface, plain TIR lens

20mm 15° frosted surface, plain TIR lens

20mm 25° microlensed surface, convex center TIR lens

20mm 30° frosted surface, convex center TIR lens

20mm 40° microlensed surface, convex center TIR lens

20mm 60° microlensed surface, convex center TIR lens

20mm 90° frosted surface, plain TIR lens

20mm 110° microlensed surface, convex center TIR lens


I see rosy beautiful tints everywhere!

Very nice reference sheet Clemence. Thanks for sharing.

Now we know we can mix high and low CCT to position the tint above/below BBL. Once Maukka’s test out, simple math like drafted by AEDe can give use relatively close approximation.


Thanks so much for posting that pic Clemence. E21A has the nicest looking tint of all emitters I’ve seen to date. Can’t wait to see Maukka’s test.

Will you make a triple MCPCB for carclo 105xx or a quad MCPCB for carclo 106xx if the E21A makes a decent beam with those TIR series ?

Triple yes, quad not yet


Thanks Clemence!! Awesome work.

I just think you missed one an important one… 353+573 :stuck_out_tongue:

Should be nearly identical to sw45k. As it stands the 403+573 is very close, but a tad cool and more neutral Duv. 453 alone is probably the closest on my monitor and in my eyes…

Sorry, I only took shots based on the four board combinations I sent to Maukka. I like the E21A sm453 R9080 because its relatively neutral yet doesn’t have too much pink as in sw45k. More shots later.
I encourage people to post here. I remember Ryzbor made some nice shots too.


I’m going to write my observations of the 3000K, 3500K mix and 4000K (d240) E21A’s but I haven’t borrowed the 3000K yet. Very shorts speaking the 3000K is nice and slightly reddish, the 3500K mix is glorious and (not overly) rosy, the d240 4000K sucks by being greenish due the too high flux bin (D220 bin will be fine) but looks extremely natural and very sunlight-like. I’m trying to talk one of my “wizard” friends to take the 4500K. Maybe Boc McBob will post his measurements here?
Al pictures are of my 3500K mix, taken with a LG G6 phone which overemphasises the rosiness.

Insemination laboratory in a abandoned manor.

3500K E21A vs 4000K XP-L2 Elf C2, look at the pipe at the ceiling, barite mine in Czech

here you see the sweet R9 popping out, partially abandoned power plant Greater Poland

made 4fun, both tints are totally wrong

my favourite tint shot place, you see everything you need to know, abandoned power plant

when someone says CRI doesn’t matter show him/her this:

[short 3500K impression here]

@Ryzbor :+1:

I bought an HDS high CRI (with Seoul SSC P4) in 2010 after seeing this post on CPF comparing CRI70 and CRI90+
Since then, for short and medium range I’m only buying high CRI flashlights and swapping LED for those with lower CRI as soon as possible. Why would you like to be “color-blinded” or suffer visual colors deficiency because of a light source ?

I modded one HDS rotary for a customer from XPG2 to Nichia 219B 9080 4500K,
guess he loved that tint and the lumen loss was very low

Great shots. Few days ago, I modded two lights, the Sofirn SP33 and the Nightwatch IRA with the 2x2000k w/ 2x4000k D240 bin E21A emitters and resulted in about 3300k cct. The tint is gorgeous. Best 3000k I’ve ever seen.

Much rosier then the 3500K mix SKV89? Lumens?
And could you tell something more about that mod? Maybe with pictures :wink:

I really would like to find a decent high-power host for the E21A. It’s technically possible to mod a Meteor with 12 E21A’s with 1.5A per single led using 20mm tir creating a flooder or with triple led setup and the stock Carclo optics with 2A per single led.

Does the E21A give a good beam profile when used with Carclo quad or triple optics?

The lights are modded with the VR21SP4 quadtrix from Virence.com. I never tried Carclo quad or triple optics. All quadtrix requires DC-Fix to eliminate holes and artifacts. I believe one of the lights I modded gave an almost artifact free beam without DC-Fix but I can’t remember exactly which one. It might be the Nightwatch IRA. The Olight R50 with the quadtrix required two layers of DC-Fix to smooth out the beam but it is well worth it. I use it as my table lamp now because the tint and CRI is so beautiful.

I mean, is the tint much rosier then a regular 1000K difference mix?
I thought about modding the Thrunite TC20 or a Convoy C8+ with a OP + DC fix but Clemence suggested this might still give a donut hole.

Next idea is to mod the upcoming 21700 S2. I could go for a quad setup with a 60dg tir and drive it at 8A but this would mean I’ll be pairing up two flooders hsince I already have a 1200lm E21A Wizard and adding a 1800lm S2 with the same optic has little sense.

So another option would be to use the VR20SP3 (?) at 6A with the Carclo triple tir. Clemence, do you know if the E21A works good with Carclo tirs? having a non-flooder E21A light with a more general-purpose beam with some range would be nice.

Since the E21A is rosier to begin with, mixing it will yield in rosier tint than your standard 1000k difference mix using other emitters. The TC20 might be hard to use with the VR16SP4 because the aluminum optic will be sitting over the wires of the 20mm mcpcb. I’m also planning to mod the 21700 S2+ with E21A and I have a bunch of Tiara Pros with E21A. I don’t think it is redundant since you can try out different tints and color temperatures.

Got the E21A & E17A boards from Clemence today. Will report next week, but here’s a teaser from 2000K+6500K tint mixing with the included optic, which is superb.

Drive current 700mA per LED. Yes, it’s very pink, but that’s what people prefer in studies. Also the Rg value is very high, which is also preferred even if not hifi per se.

Nice and even hotspot with surprisingly good definition.

Great news :smiley: . You will make the people of the Virence 4000K D240 Wizard Pro group buy I organized very happy with the measurements of the mixed 4000K and pure 4500K. 3/4 want to change the current setup…

Could you explain in normal word how high RG works? My 3500K mix has a RG103 and despite having lower CRI than the 4000K (RG100) is makes the colours “pop” noticeably more. The contrast between colours is also enhanced. I don’t know if this is caused by the higher RG or the combination od pink tint and warmer CCT. Browns (rust), reds, orange and yellow are really intensive under the 3500K mix.

bob_mcbob, will you publish your measurements of the 4500K Wizard?

I don’t have an optic in my sm453 Wizard to take measurements right now, but I do have some results I sent while informally testing it.

Main 3:
CCT = 4546K (Duv –0.0033)
Color Rendering Index (Ra) = 97.1 [ R9 = 85.7 ]

From moonlight to Main 3, Duv went from –0.0013 to –0.0033, then Turbo 1 –0.0052, Turbo 2 –0.0084. It’s below BBL on all levels, and quite rosy on turbo.

Thanks a lot Bob, this was very helpful! I think my friend is a moron. He wants to make his 4000K D240 Wizard less green by swapping one led to a 5000K D220 instead of switching to pure 4500K or making a 2x4000K + 2x4500K mix…